Distinct expression patterns of Notch ligands, Dll1 and Dll4, in normal and inflamed mice intestine

Mice

C57BL/6J mice at 6–8 weeks of age were purchased from Japan Clea (Tokyo, Japan). Lgr5- EGFP- ires- CreERT2 mice (Stock No. 008875) and R26R-lacZ mice (Stock No. 003309) were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). Dll1-floxed mice (Hozumi et al., 2004) and Dll4-floxed mice (Hozumi et al., 2008) have been previously described. To generate Lgr5- EGFP- ires- CreERT2/ Dll1^fl/fl mice or Lgr5- EGFP- ires- CreERT2/ Dll4^fl/fl mice, each floxed mice were crossed with Lgr5- EGFP- ires- CreERT2 mice. Mice carrying the R26R-LacZ allele served as control (Lgr5- EGFP- ires- CreERT2/ R26R-LacZ). Those mice were housed in the animal facility of Tokyo Medical and Dental University. All animal experiments were approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (Approval Number 0140053A).

Induction of Cre-mediated recombination

Induction of Cre-mediated gene recombination was induced by intraperitoneal injection of Tamoxifen (2 mg/body; SIGMA-ALDRICH, Missouri, USA) dissolved in corn oil for 5 consecutive days. Those mice were sacrificed at the indicated days after the first injection, and subjected to tissue analysis.

Induction of colitis

Induction of colitis was performed as previously described (Okayasu et al., 1990; Okamoto et al., 2009; Yui et al., 2012). Briefly, mice were fed ad libitum with 3% Dextran sodium sulfate (DSS, Ensuiko, Yokohama, Japan) for five consecutive days, followed by distilled water for another five days. Mice fed with distilled water alone served as control. Mice were subjected to tissue analysis at the end of the experimental period (Day 10).

Staining analysis of intestinal tissues

Immunostaining of mice intestinal tissues were done as previously described (Okamoto et al., 2009; Yui et al., 2012). Briefly, intestinal tissues were fixed in 4% parafolmaldehyde, and embedded in OCT compound (Sakura, Tokyo, Japan). 6 µm sections were prepared for staining. Primary antibodies used in the present study are summarized in Table 1. Microwave treatment (500 W, 10 min) in 10 mM citrate buffer was required for staining Dll1, Dll4, Hes1 and ATOH1. Tyramide signal amplification (Molecular Probes, California, USA) was used for immunofluorescent detection of Dll1, Dll4, Hes1 and ATOH1. Staining was visualized by secondary antibodies or Tyramide substrates conjugated with Alexa-594 or Alexa-488 (Molecular Probes, California, USA). Data were collected using a conventional epifluorecent microscope (BZ-8000 or BZ-X700, KEYENCE, Tokyo, Japan) or a confocal fluorescent microscope (FLUOVIEW FV10i; OLYMPUS, Tokyo, Japan). LacZ staining of the intestinal tissues were done as previously described (Barker et al., 2007).

Quantification of the immunostaining

Number of positive staining cells was quantified by analyzing at least 100 crypts or villi that were randomly selected from 3 individual mice. Two independent investigators, who were not informed of the origin of the tissues or the targets of the staining, collected the data. Those data were statistically analyzed with Welch’s t-test.

Dll1+ve and Dll4+ve IECs show distinct distribution pattern within the small intestinal and colonic epithelium

As previous studies have suggested that mRNA of Dll1 might be expressed within the mice intestinal epithelium (Schröder & Gossler, 2002; Crosnier et al., 2005; Stamataki et al., 2011; van Es et al., 2012), we sought to establish a method to identify cells that endogenously express Dll1 or Dll4 protein at their cell membrane, and thereby function as Notch signal-sending cells. By using a specific antibody for Dll1 or Dll4, we found that a positive staining can be yielded within the intestinal epithelium of the mice small intestine and the colon (Figs. 1A and 1D).

The specificity of those antibodies were validated and affirmed by conditional knockout of Dll1 or Dll4. By using mice carrying either the Dll1-floxed allele or the Dll4-floxed allele in addition to LGR5-EGFP-ires-CreERT2 allele, we induced genetic deletion of either Dll1 or Dll4 in LGR5-EGFP positive intestinal stem cells, and also in their progenies. In those knockout tissues, a clear abolishment of the positive staining in the progenies of EGFP-positive intestinal stem cells was confirmed in a gene-specific manner (Fig. S1).

Positive staining of both Dll1 and Dll4 showed a membranous pattern confirming that the staining demonstrates proteins that are bound to the membrane (Figs. 1B and 1E). However, the distribution of those cells along the crypt-villus axis showed a clear difference between Dll1-positive (Dll1+ve) and Dll4-positive (Dll4+ve) IECs. In the small intestine, Dll1+ve IECs resided exclusively within the crypts, but were never found in the villi (Fig. 1B). In sharp contrast, Dll4+ve IECs were found both in the villi and in the crypts. Quantitative analysis revealed that Dll1+ve IECs predominantly reside in the crypt, whereas Dll4+ve IECs mostly reside at the villus (Fig. 1C). Such a difference in distribution between Dll1+ve and Dll4+ve IECs were also found in the colon (Figs. 1D and 1E). In the colon, Dll1+ve cells were found predominantly at the lower half of the crypt, but were rarely found at the upper part of the crypt, including the surface epithelium (Fig. 1E). In contrast, Dll4+ve cells were found both in the upper and lower part of the crypt, and also at the surface epithelium. Quantitative analysis confirmed that Dll1+ve IECs reside predominantly in the lower part of the crypt, whereas Dll4+ve IECs reside both in the lower and upper part of the crypt (Fig. 1F). These results indicated that Dll1+ve and Dll4+ve IECs represent a distinct population of cells within the intestinal epithelium.

Both Dll1+ve and Dll4+ve IECs are ATOH1-positive cells, and locate adjacent to Hes1-positive IECs in the small intestinal- and colonic-crypts

Reports have suggested that Hes1 expression in IECs indicate Notch-activation and commitment to absorptive lineage cells, whereas ATOH1 expression in IECs conversely indicates lack of Notch activation and commitment to secretory lineage cells (van Den Brink, de Santa Barbara & Roberts, 2001; Vooijs, Liu & Kopan, 2011). To identify their potential to serve as Notch activating cells, and gain insight into their cell characteristics, we conducted a double immunostaining analysis of Dll1 or Dll4 with Hes1 or ATOH1. In the small intestinal crypts, both Dll1+ve and Dll4+ve cells were found adjacent to Hes1-positive (Hes1+ve) cells (Figs. 2A and 2B). In the villi, the adjacent cells of Dll4+ve cells were never found to be positive for Hes1. Also, both Dll1+ve and Dll4+ve IECs themselves were never found to be positive for Hes1. Instead, both Dll1+ve and Dll4+ve IECs were invariably positive for ATOH1 (Fig. 2B). Exactly the same expression patterns of Hes1 and ATOH1 within Dll1+ve and Dll4+ve IECs, or in their neighboring cells were found in the colonic crypts (Figs. 2C and 2D). A horizontal cross section of the crypt region further confirmed those expression patterns of Hes1 and ATOH1 in Dll1+ve and Dll4+ve IECs (Fig. S2). These findings clearly illustrate the definite site of lateral inhibition by ATOH1-positive (ATOH1+ve) crypt cells, through their expression of Dll1 or Dll4. In contrast, those Dll4+ve IECs that were found at the small intestinal villi or at the colonic surface epithelium appeared to co-express ATOH1, but were never found adjacent to Hes1+ve cells, indicating that those IECs may not function as Notch activating cells under normal conditions.

Dll1+ve and Dll4+ve IECs represent distinct populations of secretory lineage cells in the small intestine and in the colon

To further identify the properties of Dll1+ve or Dll4+ve IECs, we further checked the co-expression of a proliferative cell marker, Ki67. Consistent with the former report (Stamataki et al., 2011), co-immunostaining with Ki67 showed that both Dll1+ve and Dll4+ve IECs are mostly negative for Ki67, indicating that they are post-mitotic in the small intestine (Fig. 3A) and in the colon (Fig. 3B). Quantitative analysis showed that 34.5% of Dll1+ve IECs and 19.3% of Dll4+ve IECs in the small intestine co-express Ki-67 (Table S1). Also, 22.6% of Dll1+ve IECs and 14.8% of Dll4+ve IECs in the colon co-expressed Ki-67. However, those values varied among individual crypts. As our former results suggested that both Dll1+ve and Dll4+ve IECs are ATOH1+ve cells that are committed to the secretory cell lineage, we analyzed the expression of secretory lineage differentiation markers in Dll1+ve or Dll4+ve IECs. In the small intestine, both Dll1+ve and Dll4+ve goblet cells were clearly present (Fig. 4A). Also, a few number of Dll4+ve enteroendocrine cells were found (Table S1). However, enteroendocrine cells that are positive for Dll1, or tuft cells that are positive for either Dll1 or Dll4, were never observed. In the colon, both Dll1+ve and Dll4+ve goblet cells were also clearly present (Fig. 4B). However, enteroendocrine cells or tuft cells that are positive for Dll1 or Dll4 were never observed. Quantitative analysis showed that increased number of Dll4+ve IECs co-express MUC2, compared to Dll1+ve IECs, both in the small intestine and in the colon (Table S1). Thus, the results indicated that both Dll1+ve and Dll4+ve IECs could acquire goblet cell phenotype, but rarely exhibit the phenotype of other secretory lineage cells. However, as compared to Dll1+ve IECs, Dll4+ve IECs appeared to include higher number of terminally differentiated goblet cells and can additionally acquire enteroendocrine cell phenotype, which confirmed the idea that Dll1+ve and Dll4+ve IECs represent distinct population of secretory lineage cells.

Dll1+ve or Dll4+ve IECs constitute the intestinal stem cell niche

Previous studies have suggested that Paneth cells express Dll1 or Dll4, and thereby constitute the key component of the small intestinal stem cell niche (Stamataki et al., 2011; Sato et al., 2012). However, endogenous protein expression of Dll1 or Dll4 in those stem cell niche cells has never been demonstrated. So far, our immunohistochemical analysis of small intestinal tissues failed to detect expression of Dll1 in any Paneth cell (Fig. 5A). However, a clear expression of Dll4 was found in a very rare population of Paneth cells (Fig. 5B). In the colon, it has been shown that c-kit positive cells take the place of Paneth cells, and constitute the colonic stem cell niche (Rothenberg et al., 2012). Double immunostaining with c-kit clearly showed that Dll1+ve IECs mostly overlap with the c-kit-positive cell population in the colon (Fig. 5C). Also, those Dll4+ve IECs that reside at the lower part of the colonic crypts clearly co-expressed c-kit (Fig. 5D). Quantitative analysis showed that almost all Dll1+ve IECs are c-kit positive, whereas around 70% of Dll4+ve IECs are c-kit positive in the colon (Table S1). Thus, our analysis showed that Dll1+ve IECs or Dll4+ve IECs surely reside as a component of the stem cell niche in small intestinal and colonic crypts.

On the other hand, our analysis of Lgr5-EGFP-ires-CreERT2 mice showed that Dll1+ve IECs or Dll4+ve IECs are mostly negative for LGR5-EGFP, suggesting that those cell populations do not include the LGR5-positive stem cell itself. However, we found a rare population of Dll1+ve or Dll4+ve IECs that co-express LGR5-EGFP, at the upper end of the stem cell region (Fig. S3). Those cells appeared to express EGFP at a relatively low level, and preferentially located at the +5 position. Thus these Dll1+EGFP+double positive cells, or Dll4+EGFP+double positive cells may represent the bi-potent secretory progenitor cells as described previously by van Es et al. (2012).

Dll4+ve cells dominate the colonic crypts of the DSS-colitis mice

Our previous study has shown that number of Hes1+ve cells significantly increase in the colitic mucosa of DSS-colitis mice, and also in ulcerative colitis patients, indicating that the canonical Notch pathway is activated in those cells (Okamoto et al., 2009; Zheng et al., 2011). However, the ligand that is responsible for such an increase in Notch activation remains uncertain. Therefore, we examined the expression of Dll1 and Dll4 in tissues of DSS-colitis mice. Analysis showed a surprising loss of Dll1 expression in the crypts of DSS-colitis mice (Fig. 6A). Conversely, a striking increase in the number of Dll4+ve IECs was observed in the crypts of those DSS-colitis mice, suggesting that a distinct regulation of Dll1 and Dll4 expression exists under the inflammatory environment. Quantitative analysis confirmed that number of Dll1+ve IECs significantly decrease, whereas that of Dll4+ve IECs significantly increase in DSS-colitis mice, compared to control mice (Fig. 6B). Double staining with Hes1 confirmed that a rare population of Dll1+ve IECs, as well as the dominant Dll4+ve IECs clearly located adjacent to Hes1+ve IECs within the elongated crypts of the DSS-colitis mice (Fig. 6C). Of note, those Hes1+ve IECs appearing at the mid-to-upper part of the colitic crypts were mostly accompanied adjacently by Dll4+ve IECs. Also, the analysis showed that both Dll1+ve and Dll4+ve IECs that were found in the DSS-colitis mice maintain their expression of ATOH1 (Fig. 6D). A horizontal cross section of the lower and upper crypt region further confirmed those expression patterns of Hes1 and ATOH1 in Dll4+ve IECs (Fig. S4). Thus, results indicated that the Dll1- or Dll4-ligand mediated Notch activation, as well as the lateral inhibition system, is maintained and present also in the inflamed colonic crypts.